Not normally distributed errors after controlling for time fixed effects

[ana maria giraldo]

Hello everyone, I am running a regression where I would like to control for both region and time fixed effects.

When I only control for region fixed eff my errors are normally distributed. But as soon as I control for time fixed effects errors are not nor distributed anymore.
I think in my case it is important to meet this assumption of ols because I have to do hypothesis testing. Please find my results attached.


I would like to know the reasons why this might be the case and how could I solve it ( I am aware that I could use xtreg for fixed effects in panel data, but I would like to continue working with dummy variables)
I already transformed my explanatory var to make them normal, I also tried the reggresion with lnyrsc instead of yrsc, but nothing worked so far


Thanks for your help

[Carlo Lazzaro]

Ana Maria:
you might be interested in the following thread https://www.statalist.org/forums/for...kewness-part-2.
Besides:
- I'm not clear with your using -regress- instead of -xtreg,fe- with panel data;
- I would not sponsor using -xi:- prefix and recommend to create categorical variables and interaction via -fvvarlist- notation. instead